INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 panels

-0.81

 suicides

-0.69

é¾į

-0.68

 assemblies

-0.67

 levels

-0.62

 bins

-0.61

 ratios

-0.60

 panel

-0.58

 joints

-0.58

 ratio

-0.58

POSITIVE LOGITS

 until

1.14

until

1.10

 till

0.95

 Until

0.86

govtrack

0.85

SHIP

0.81

TRY

0.80

hower

0.79

Until

0.79

hovah

0.78

Activations Density 0.000%

No Known Activations

This feature has no known activations.